The other thing is that too often when some industry pundits are looking at disk versus. tape, they're looking at a price curve for acquisition, not the total cost of ownership characteristics -- the electricity to keep disk spinning, the floor space requirements, the personnel requirements. The TCO advantage of tape for long-term retention, coupled with the advantages of WORM tape for compliance purposes, have kept tape not only relevant but very cost effective for our clients in a tiered-storage, long-term data retention environment.
Andrews: A couple of points: People tend to have trouble understanding the price scale properly. Disk ends up being pennies per gigabyte and tape a fraction of a penny, and those seem like small differences. Maybe they would be, if you were storing just a gigabyte. The problem is that the relevant numbers to look at aren't with the absolute dollar -- it's the overall relative relationship between the cost of disk and the cost of tape. The growth of data and the growth of retention requirements that'll drive even more data mean that you might have, say, 10 petabytesof data. Half a cent difference between the cost of disk and the cost of tape then turns out to be a lot of money.
Our view is that if you really want to have an effective ILM management infrastructure, then you need to have tape in that infrastructure because it gives you so much cost leverage. There's a ton of data you're going to have archived and with the long-term data you don't access a lot, you don't need to have it on disks up and spinning. So not only is acquisition cost higher, but the maintenance cost is higher. When you do the cost benefit analysis then, tape still has a large contribution to make.
Your SAN Volume Controller (SVC) product was originally released in two versions -- as an appliance and as a switch. Now it seems the switch version is being taken off the market. Veritas has also discontinued a similar product. Why is this? Why is it harder to sell switch-based virtualization?Lechner: We have not removed the switch-based version from the market, but we have been more successful to date with the appliance version of SVC. Have you seen EMC's Invista in the market yet? What are SVC's technical merits versus Invista? Lechner: We have not seen EMC's product in the market yet … Because SVC is an appliance, its reach across the market is greater. We are seeing rapid adoption of SVC by clients in SMBs (<999 employees), which is a market that EMC has chosen to completely ignore. Invista requires a specific set of intelligent switches, which are new to the market so are not broadly integrated into current infrastructures. It has a different cost profile than would align with smaller customers. Finally, SVC is on its seventh generation in the market, so has a track record of reliability and stability as well as a very broad set of supported environments -- all of which are critical for customer implementations of virtualization. Return to Page 1